jonatasgrosman's picture
Update README.md
489b34f
metadata
language: el
datasets:
  - common_voice
metrics:
  - wer
  - cer
tags:
  - audio
  - automatic-speech-recognition
  - speech
  - xlsr-fine-tuning-week
license: apache-2.0
model-index:
  - name: XLSR Wav2Vec2 Greek by Jonatas Grosman
    results:
      - task:
          name: Speech Recognition
          type: automatic-speech-recognition
        dataset:
          name: Common Voice el
          type: common_voice
          args: el
        metrics:
          - name: Test WER
            type: wer
            value: 11.62
          - name: Test CER
            type: cer
            value: 3.36

Fine-tuned XLSR-53 large model for speech recognition in Greek

Fine-tuned facebook/wav2vec2-large-xlsr-53 on Greek using the train and validation splits of Common Voice 6.1 and CSS10. When using this model, make sure that your speech input is sampled at 16kHz.

This model has been fine-tuned thanks to the GPU credits generously given by the OVHcloud :)

The script used for training can be found here: https://github.com/jonatasgrosman/wav2vec2-sprint

Usage

The model can be used directly (without a language model) as follows...

Using the HuggingSound library:

from huggingsound import SpeechRecognitionModel

model = SpeechRecognitionModel("jonatasgrosman/wav2vec2-large-xlsr-53-greek")
audio_paths = ["/path/to/file.mp3", "/path/to/another_file.wav"]

transcriptions = model.transcribe(audio_paths)

Writing your own inference script:

import torch
import librosa
from datasets import load_dataset
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor

LANG_ID = "el"
MODEL_ID = "jonatasgrosman/wav2vec2-large-xlsr-53-greek"
SAMPLES = 5

test_dataset = load_dataset("common_voice", LANG_ID, split=f"test[:{SAMPLES}]")

processor = Wav2Vec2Processor.from_pretrained(MODEL_ID)
model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)

# Preprocessing the datasets.
# We need to read the audio files as arrays
def speech_file_to_array_fn(batch):
    speech_array, sampling_rate = librosa.load(batch["path"], sr=16_000)
    batch["speech"] = speech_array
    batch["sentence"] = batch["sentence"].upper()
    return batch

test_dataset = test_dataset.map(speech_file_to_array_fn)
inputs = processor(test_dataset["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)

with torch.no_grad():
    logits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits

predicted_ids = torch.argmax(logits, dim=-1)
predicted_sentences = processor.batch_decode(predicted_ids)

for i, predicted_sentence in enumerate(predicted_sentences):
    print("-" * 100)
    print("Reference:", test_dataset[i]["sentence"])
    print("Prediction:", predicted_sentence)
Reference Prediction
ΤΟ ΒΑΣΙΛΌΠΟΥΛΟ, ΠΟΥ ΜΟΙΆΖΕΙ ΛΕΟΝΤΑΡΆΚΙ ΚΑΙ ΑΕΤΟΥΔΆΚΙ ΤΟ ΒΑΣΙΛΌΠΟΥΛΟ ΠΟΥ ΜΙΑΣΕ ΛΙΟΝΤΑΡΑΚΉ ΚΑΙ ΑΪΤΟΥΔΆΚΙ
ΣΥΝΆΜΑ ΞΕΠΡΌΒΑΛΑΝ ΑΠΌ ΜΈΣΑ ΑΠΌ ΤΑ ΔΈΝΤΡΑ, ΔΕΞΙΆ, ΑΡΜΑΤΩΜΈΝΟΙ ΚΑΒΑΛΑΡΈΟΙ. ΣΥΝΆΜΑ ΚΑΙ ΤΡΌΒΑΛΑΝ ΑΠΌ ΜΈΣΑ ΑΠΌ ΤΑ ΔΈΝΤΡΑ ΔΕΞΙΆ ΑΡΜΑΤΩΜΈΝΟΙ ΚΑΒΑΛΑΡΈΟΙ
ΤΑ ΣΥΣΚΕΥΑΣΜΈΝΑ ΒΙΟΛΟΓΙΚΆ ΛΑΧΑΝΙΚΆ ΔΕΝ ΠΕΡΙΈΧΟΥΝ ΣΥΝΤΗΡΗΤΙΚΆ ΚΑΙ ΟΡΜΌΝΕΣ ΤΑ ΣΥΣΚΕΦΑΣΜΈΝΑ ΒΙΟΛΟΓΙΚΆ ΛΑΧΑΝΙΚΆ ΔΕΝ ΠΕΡΙΈΧΟΥΝ ΣΙΔΗΡΗΤΙΚΆ ΚΑΙ ΟΡΜΌΝΕΣ
ΑΚΟΛΟΥΘΉΣΕΤΕ ΜΕ! ΑΚΟΛΟΥΘΉΣΤΕ ΜΕ
ΚΑΙ ΠΟΎ ΜΠΟΡΏ ΝΑ ΤΟΝ ΒΡΩ; Ε ΠΟΎ ΜΠΟΡΏ ΝΑ ΤΙ ΕΒΡΩ
ΝΑΙ! ΑΠΟΚΡΊΘΗΚΕ ΤΟ ΠΑΙΔΊ ΝΑΙ ΑΠΟΚΡΊΘΗΚΕ ΤΟ ΠΑΙΔΊ
ΤΟ ΠΑΛΆΤΙ ΜΟΥ ΤΟ ΠΡΟΜΉΘΕΥΕ. ΤΟ ΠΑΛΆΤΙ ΜΟΥ ΤΟ ΠΡΟΜΉΘΕΥΕ
ΉΛΘΕ ΜΉΝΥΜΑ ΑΠΌ ΤΟ ΘΕΊΟ ΒΑΣΙΛΙΆ; ΉΛΘΑ ΜΕΊΝΕΙ ΜΕ ΑΠΌ ΤΟ ΘΕΊΟ ΒΑΣΊΛΙΑ
ΠΑΡΑΚΆΤΩ, ΈΝΑ ΡΥΆΚΙ ΜΟΥΡΜΟΎΡΙΖΕ ΓΛΥΚΆ, ΚΥΛΏΝΤΑΣ ΤΑ ΚΡΥΣΤΑΛΛΈΝΙΑ ΝΕΡΆ ΤΟΥ ΑΝΆΜΕΣΑ ΣΤΑ ΠΥΚΝΆ ΧΑΜΌΔΕΝΤΡΑ. ΠΑΡΑΚΆΤΩ ΈΝΑ ΡΥΆΚΙ ΜΟΥΡΜΟΎΡΙΖΕ ΓΛΥΚΆ ΚΥΛΏΝΤΑΣ ΤΑ ΚΡΥΣΤΑΛΛΈΝΙΑ ΝΕΡΆ ΤΟΥ ΑΝΆΜΕΣΑ ΣΤΑ ΠΥΚΡΆ ΧΑΜΌΔΕΝΤΡΑ
ΠΡΆΓΜΑΤΙ, ΕΊΝΑΙ ΑΣΤΕΊΟ ΝΑ ΠΆΡΕΙ Ο ΔΙΆΒΟΛΟΣ ΠΡΆΓΜΑΤΗ ΕΊΝΑΙ ΑΣΤΕΊΟ ΝΑ ΠΆΡΕΙ Ο ΔΙΆΒΟΛΟΣ

Evaluation

The model can be evaluated as follows on the Greek test data of Common Voice.

import torch
import re
import librosa
from datasets import load_dataset, load_metric
from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor

LANG_ID = "el"
MODEL_ID = "jonatasgrosman/wav2vec2-large-xlsr-53-greek"
DEVICE = "cuda"

CHARS_TO_IGNORE = [",", "?", "¿", ".", "!", "¡", ";", ";", ":", '""', "%", '"', "�", "ʿ", "·", "჻", "~", "՞",
                   "؟", "،", "।", "॥", "«", "»", "„", "“", "”", "「", "」", "‘", "’", "《", "》", "(", ")", "[", "]",
                   "{", "}", "=", "`", "_", "+", "<", ">", "…", "–", "°", "´", "ʾ", "‹", "›", "©", "®", "—", "→", "。",
                   "、", "﹂", "﹁", "‧", "~", "﹏", ",", "{", "}", "(", ")", "[", "]", "【", "】", "‥", "〽",
                   "『", "』", "〝", "〟", "⟨", "⟩", "〜", ":", "!", "?", "♪", "؛", "/", "\\\\", "º", "−", "^", "ʻ", "ˆ"]

test_dataset = load_dataset("common_voice", LANG_ID, split="test")

wer = load_metric("wer.py") # https://github.com/jonatasgrosman/wav2vec2-sprint/blob/main/wer.py
cer = load_metric("cer.py") # https://github.com/jonatasgrosman/wav2vec2-sprint/blob/main/cer.py

chars_to_ignore_regex = f"[{re.escape(''.join(CHARS_TO_IGNORE))}]"

processor = Wav2Vec2Processor.from_pretrained(MODEL_ID)
model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)
model.to(DEVICE)

# Preprocessing the datasets.
# We need to read the audio files as arrays
def speech_file_to_array_fn(batch):
    with warnings.catch_warnings():
        warnings.simplefilter("ignore")
        speech_array, sampling_rate = librosa.load(batch["path"], sr=16_000)
    batch["speech"] = speech_array
    batch["sentence"] = re.sub(chars_to_ignore_regex, "", batch["sentence"]).upper()
    return batch

test_dataset = test_dataset.map(speech_file_to_array_fn)

# Preprocessing the datasets.
# We need to read the audio files as arrays
def evaluate(batch):
    inputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)

    with torch.no_grad():
        logits = model(inputs.input_values.to(DEVICE), attention_mask=inputs.attention_mask.to(DEVICE)).logits

    pred_ids = torch.argmax(logits, dim=-1)
    batch["pred_strings"] = processor.batch_decode(pred_ids)
    return batch

result = test_dataset.map(evaluate, batched=True, batch_size=8)

predictions = [x.upper() for x in result["pred_strings"]]
references = [x.upper() for x in result["sentence"]]

print(f"WER: {wer.compute(predictions=predictions, references=references, chunk_size=1000) * 100}")
print(f"CER: {cer.compute(predictions=predictions, references=references, chunk_size=1000) * 100}")

Test Result:

In the table below I report the Word Error Rate (WER) and the Character Error Rate (CER) of the model. I ran the evaluation script described above on other models as well (on 2021-04-22). Note that the table below may show different results from those already reported, this may have been caused due to some specificity of the other evaluation scripts used.

Model WER CER
lighteternal/wav2vec2-large-xlsr-53-greek 10.13% 2.66%
jonatasgrosman/wav2vec2-large-xlsr-53-greek 11.62% 3.36%
vasilis/wav2vec2-large-xlsr-53-greek 19.09% 5.88%
PereLluis13/wav2vec2-large-xlsr-53-greek 20.16% 5.71%

Citation

If you want to cite this model you can use this:

@misc{grosman2021xlsr53-large-greek,
  title={Fine-tuned {XLSR}-53 large model for speech recognition in {G}reek},
  author={Grosman, Jonatas},
  howpublished={\url{https://huggingface.co/jonatasgrosman/wav2vec2-large-xlsr-53-greek}},
  year={2021}
}